Research on Dynamic Programming Strategy of Bayesian Network Structure Learning

Table 5

Accuracy comparison between three algorithms under standard network data sets.

BNs	Nodes	Edges	Tns	Sample sizes	DFSDP	CBDP	BFSDP
Weather	4	4	8	1000	−984.7736	−984.7736	−984.7736
				2000	−3.79e + 003	−3.79e + 003	−3.79e + 003
				5000	−9.57e + 003	−9.57e + 003	−9.57e + 003
				10000	−1.91e + 004	−1.91e + 004	−1.91e + 004
Cancer	5	4	10	1000	−1.04e + 003	−1.04e + 003	−1.04e + 003
				2000	−4.18e + 003	−4.18e + 003	−4.18e + 003
				5000	−1.04e + 004	−1.04e + 004	−1.04e + 004
				10000	−2.10e + 004	−2.10e + 004	−2.10e + 004
Earthquake	5	4	10	1000	−191.4748	−199.6867	−191.4748
				2000	−905.1324	−905.1324	−905.1324
				5000	−2.23e + 003	−2.23e + 003	−2.23e + 003
				10000	−4.42e + 003	−4.42e + 003	−4.42e + 003
Survey	6	6	14	1000	−2.00e + 003	−2.00e + 003	−2.00e + 003
				2000	−8.05e + 003	−8.05e + 003	−8.05e + 003
				5000	−2.00e + 004	−2.00e + 004	−2.00e + 004
				10000	−3.95e + 004	−3.95e + 004	−3.95e + 004
Asia	8	8	16	1000	−1.10e + 003	−1.11e + 003	−1.10e + 003
				2000	−4.55e + 003	−4.55e + 003	−4.55e + 003
				5000	−1.11e + 004	−1.11e + 004	−1.11e + 004
				10000	−2.25e + 004	−2.25e + 004	−2.25e + 004
Sachs	11	17	33	1000	−3.87e + 003	−5.68e + 003	−3.87e + 003
				2000	−1.46e + 004	−1.83e + 004	−1.46e + 004
				5000	−3.60e + 004	4.12e + 004	−3.60e + 004
				10000	−7.20e + 004	8.00e + 004	−7.20e + 004